A New Wrapper-Based Feature Selection Technique with Fireworks Algorithm for Android Malware Detection

Abstract

Smartphone use has expanded dramatically in recent years, particularly for Android-based smartphones, due to their wide availability and competitive pricing compared to non-Android devices. The significant increase in the use of Android applications has resulted in a spike in the number of malicious applications, which represent a severe danger to user privacy. In this paper, the authors proposed FWA-FS, a novel method for Android malware detection with feature selection based on the fireworks algorithm. Static analysis is used in the proposed technique to classify applications as benign or malicious. To describe applications, they employ permissions derived from APK files as feature vectors. The most important features were then chosen using the proposed FWA-FS method. Finally, to develop classification models, different machine learning algorithms were trained using specified features. According to experimental findings, the suggested strategy can greatly enhance classification performance with an average increase of 6% and 25% in accuracy for KNN and Naïve Bayes respectively.